Skip to main content

APC UPS - Graceful Shutdown for AHV

  • July 19, 2017
  • 7 replies
  • 2116 views

Forum|alt.badge.img+4
I am currently running VMware but will be soon switching to AHV.

Will be purchasing a APC UPS with parchute software.

Does anyone know if this is compatible for AHV?

Does anyone out there with a UPS running AHV care to comment how they are managing the graceful shutdown of their Nodes and VMs in the event of power loss?
This topic has been closed for replies.

7 replies

Forum|alt.badge.img+8
  • Nutanix Employee
  • July 21, 2017
Hello,

I reached out to our alliances team to confirm what the current status on this is.

APC Powerchute is in dev stage right now for supporting graceful shutdown of AHV clusters ... we currently lack APIs for shutting down cluster so they are evaluating using script/cli to trigger a shutdown ... we are currently working with them on this.

Forum|alt.badge.img+4
  • Author
  • Adventurer
  • July 24, 2017
Thanks for the response Brian.

With regards to the scripting side, did you say this is avaialble right now?

My reseller is looking into using this as workround but I am not if this is feasible at this stage.

Forum|alt.badge.img+8
  • Nutanix Employee
  • July 24, 2017
Hello,

The scripting option is still work in progress and not available yet.

Forum|alt.badge.img+4
  • Voyager
  • November 14, 2017
Any update on the status of this? Thanks.

I would like if Nutanix made progress on that.

  • Voyager
  • March 9, 2018
Has there been any traction on this issue? We have many ROBO sites with a ESXi cluster fed from an APC UPS. I would like to have a cluster shutdown command initiated and the CVMs shut down when on battery for x minutes. Currently we have PowerChute installed on the guests that gracefully shutdown, but I would like for the cluster to go down gracefully as well.

  • Voyager
  • June 8, 2018
IIRC PowerChute could execute scripts... but it would be clunky and most likely have to embed authentication information into the script and run from a VM that was itself on the cluster - causing itself to flush itself when it pulled the trigger.

Man, I would have hopped for a much more graceful option.